Mechanical arm rapid obstacle avoidance path planning method for variable scene

Abstract

The invention provides a variable scene-oriented mechanical arm rapid obstacle avoidance path planning method, which comprises the following steps of: representing a mechanical arm configuration in a joint space coordinate form, and obtaining a map sampling point set by using a map sampling point selection strategy based on a random number generator; according to mapping of joint space coordinates and Cartesian space mechanical arm tail end coordinates, a representation mode of obstacles in a road sign map is designed; according to known state change information of obstacles in the scene, a multi-type sampling point communication strategy is designed to construct a new road sign map, and the road sign map containing all change state information of the obstacles is obtained; and according to the obtained road sign map, a road sign map search strategy for a variable scene is designed based on an A * algorithm, and the mechanical arm path search process is completed. According to the technical scheme provided by the embodiment of the invention, the mechanical arm can realize quick search of the shortest path in a variable scene on the premise of avoiding collision.

Description

[Technical field]

[0001] The invention belongs to the field of robot arm obstacle avoidance motion planning, and relates to a robot arm fast obstacle avoidance path planning method for variable scenarios. [Background technology]

[0002] With the continuous advancement of automation technology, robotic arms have become key actuators in many fields such as industrial production and aerospace. However, in actual working environments, the environment in which the robotic arm is located and its joint components may limit the motion planning of the robotic arm. Therefore, when performing work tasks, the robotic arm must consider the obstacles in the robotic arm body and the environment to quickly plan a collision-free path to ensure that the task is completed safely and efficiently. Since the time-consuming obstacle avoidance path planning directly affects the work efficiency of the robotic arm, in-depth research on the rapid obstacle avoidance method of the robotic arm and improving the efficiency of obstacle avoidance path planning have great theoretical value and practical significance for promoting the widespread application of robotic arms in industrial production, aerospace and other fields.

[0003] At present, the robot arm obstacle avoidance path planning methods are mainly divided into search-based methods and sampling-based methods. The typical representative of the search-based method is the traditional heuristic search method, which requires real-time collision detection of the robot arm during the planning process, which undoubtedly increases the time cost of path planning and is difficult to meet the efficiency requirements of robot arm obstacle avoidance path planning.

[0004] The random landmark map method (PRM algorithm for short) is a typical sampling-based method that can greatly improve planning efficiency compared to traditional search-based methods. The PRM algorithm is mainly divided into two parts: the learning stage and the search stage. The learning stage takes a lot of time to complete the selection of map sampling points and the construction of the landmark map. The search stage completes the search for collision-free paths in the constructed landmark map. Since there is no need for a large number of collision detection processes at this time, the path planning time cost is low. Although the traditional PRM algorithm has high planning efficiency, it cannot cope with the situation where the working scene will change. When the state of obstacles in the scene where the robot arm is located changes, it takes a lot of time to re-build a new landmark map. Therefore, it is necessary to propose a new robot arm fast obstacle avoidance path planning method based on the random landmark map method, which can quickly complete the planning of the obstacle avoidance path under the premise of considering the changes in the working scene. [Summary of the invention]

[0005] In view of this, the present invention provides a method for fast obstacle avoidance path planning of a robot arm for variable scenes. The proposed method can solve the problem that the traditional random landmark map method cannot be applied to variable scenes, and can still efficiently and safely complete the obstacle avoidance path planning when the scene changes.

[0006] The present invention provides a method for fast obstacle avoidance path planning of a robot arm for variable scenarios, comprising:

[0007] Step S1 represents the configuration of the robot arm in the form of joint space coordinates, and uses a map sampling point selection strategy based on a random number generator to obtain a map sampling point set;

[0008] Step S2 designs a method for representing obstacles in the roadmap based on the mapping of joint space coordinates and Cartesian space robot end coordinates;

[0009] Step S3 designs a multi-type sampling point connection strategy to construct a new landmark map based on the known state change information of obstacles in the scene, and obtains a landmark map containing all the state change information of obstacles;

[0010] In step S4, a landmark map search strategy for variable scenarios is designed based on the A* algorithm according to the obtained landmark map to complete the robot arm path search process.

[0011] In the above method, step S1 comprises:

[0012] Step S1.1 represents a certain state of the k-degree-of-freedom robot in the workspace as a k-dimensional vector:

[0013] q=(θ1,θ2,...,θ k )

[0014] Among them, θ i represents the joint angle of the i-th joint of the robot;

[0015] Step S1.2 The map sampling point selection strategy based on the random number generator includes:

[0016] Step S1.2.1: Set the number of map sampling points j = 1;

[0017] Step S1.2.2: let joint angle i = 1;

[0018] Step S1.2.3 uses the random number generator rand() to randomly generate joint angles within the limits of the i-th joint angle of the robot arm. The expression is:

[0019] θ i =rand()%(θ i max -θ i min )+θ i min

[0020] Among them, θ i max ,θ i minThey represent the maximum and minimum limits of the i-th joint angle of the robot arm, respectively, and % represents the remainder operation;

[0021] Step S1.2.4 Initialize a k-dimensional zero vector q j , represents the configuration of the k-degree-of-freedom manipulator, and θ i Assign to q j The expression for the elements at the corresponding positions is:

[0022] q j [i] = θ i

[0023] Among them, q j [i] represents vector q j The element at position i;

[0024] Step S1.2.5: let i=i+1, if i≤k, return to step S1.2.3;

[0025] Step S1.2.6 obtains a set of robot arm configurations q j =(θ1,θ2,...,θ k ), the robot arm configuration q j As a sampling point of the landmark map, add it to the map sampling point set P;

[0026] Step S1.2.7: let j=j+1;

[0027] Step S1.2.8: Let the maximum number of sampling points of the map be M. If j≤M, return to step S1.2.2, otherwise skip this step;

[0028] Step S1.2.9 obtains the map sampling point set P:

[0029] P={q1,q2,…,q M}

[0030] Among them, the selection of M is set according to the task requirements.

[0031] In the above method, step S2 comprises:

[0032] Take out all the sampling points q=(θ1,θ2,...,θ k ), perform collision detection on all sampling points in turn, filter out all sampling points that collide with the robot itself or obstacles, use set form to represent the sampling points that collide with obstacles, and obtain the set of sampling points that collide with obstacles in the working environment:

[0033] S={q1,q2,...,q N}

[0034] Where S represents the set of sampling points where the robot arm collides with obstacles, q i (i=1, 2, ..., N) represents the sampling points in the set that collide with obstacles in the working environment. The sampling point set S is used to represent the obstacles in the working scene in the roadmap.

[0035] In the above method, step S3 comprises:

[0036] Step S3.1: obstacles in the workspace are divided into fixed obstacles and variable obstacles. For variable obstacles, all variable states of the obstacles are represented in the workspace.

[0037] Step S3.2 divides the existing configuration set S of the robot arm colliding with obstacles into sets S1 and S2 according to the collision with fixed obstacles and the collision with variable obstacles. S2 represents the configuration set of the robot arm colliding with various possible states of variable obstacles:

[0038]

[0039] in It represents the configuration in which the robot arm collides with variable obstacles A, B, etc. in states a, b, etc.;

[0040] Step S3.3 The strategy for connecting multiple sampling points on the roadmap includes:

[0041] Step S3.3.1: divide the obtained map sampling point set P into a collision point set S that collides with obstacles and a safe sampling point set P0 that does not collide with obstacles, and divide the collision point set S into a dangerous sampling point set S1 that collides with fixed obstacles and a threat sampling point set S2 that collides with variable obstacles;

[0042] Step S3.3.2 deletes the dangerous sampling point set S1 from the road sign map, does not participate in the connection of sampling points in the road sign map, and sets the remaining sampling point set in the road sign map to R;

[0043] Step S3.3.3: Let the number of sampling points in the road sign map i=1, and the number of sampling points in the sampling point set R be r;

[0044] Step S3.3.4 Calculate the sampling point q i The distance to all sampling points in the sampling point set R is expressed as:

[0045]

[0046] Among them, k is the degree of freedom of the robot arm, q j is the sampling point in set R, d j is the sampling point q i With sampling point qj The distance between the sampling points q i The distances from all sampling points in the sampling point set R are calculated using the set D i To characterize.

[0047] Step S3.3.5 In set D i Select the distance sampling point q i The nearest h sampling points are connected. If the sampling point to be connected belongs to the safe sampling point set P0, the corresponding connected edge is defined as a reliable edge. If the sampling point to be connected belongs to the threat sampling point set S2, the corresponding connected edge is defined as an unreliable edge.

[0048] Step S3.3.6: Let i = i + 1. If i ≤ r, return to step S3.3.4; otherwise, skip this step.

[0049] Step S3.3.7 obtains the landmark map G.

[0050] In the above method, step S4 comprises:

[0051] Step S4.1 obtains the current state of the variable obstacle from the working environment;

[0052] Step S4.2: Take out the sampling point set S2' representing all states of the variable obstacle from the threat sampling point set S2, match it with the current state of the variable obstacle, and find the sampling point q that matches the current state of the variable obstacle in the sampling point set S2' representing all states of the variable obstacle. danger ;

[0053] Step S4.3: Set the sampling point q danger Convert it into dangerous sampling points, and all other sampling points in the threat sampling point set S2 are converted into safe sampling points, and the threat sampling point set S2 in the road sign map G is converted into dangerous sampling points and safe sampling point sets;

[0054] Step S4.4 converts the unreliable edges connected to the dangerous sampling points into dangerous edges, and sets the weights to +∞; converts the unreliable edges not connected to the dangerous sampling points into reliable edges, and does not change the weights;

[0055] Step S4.5 uses the A* algorithm to search for the path with the smallest sum of edge weights from the starting point to the end point in the new landmark map, completing the robot arm's rapid obstacle avoidance path planning.

Brief Description of the Drawings

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creativity and labor.

[0057] Figure 1 It is a flow chart of a method for fast obstacle avoidance path planning of a manipulator for variable scenarios provided by an embodiment of the present invention;

[0058] Figure 2 is a schematic diagram of a four-degree-of-freedom robotic arm model used in a simulation experiment in an embodiment of the present invention;

[0059] Figure 3 is a schematic diagram of a planning result in which a variable obstacle is changed into a fixed obstacle in an embodiment of the present invention;

[0060] Figure 4 It is a schematic diagram of planning results in which a variable obstacle is not changed into a fixed obstacle in an embodiment of the present invention. [Specific embodiment]

[0061] In order to better understand the technical solution of the present invention, the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0062] It should be clear that the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] The embodiment of the present invention provides a method for fast obstacle avoidance path planning of a robotic arm for variable scenarios. Figure 1 , which is a flow chart of a method for rapid obstacle avoidance path planning of a manipulator for variable scenarios provided by an example of the present invention, such as Figure 1 As shown, the method comprises the following steps:

[0064] Step 101, characterizing the configuration of the manipulator in the form of joint space coordinates, and using a map sampling point selection strategy based on a random number generator to obtain a map sampling point set;

[0065] Specifically, firstly, a certain state of the k-degree-of-freedom robot in the workspace is represented as a k-dimensional vector:

[0066] q=(θ1,θ2,…,θ k )

[0067] Among them, θ i represents the joint angle of the i-th joint of the robot;

[0068] Secondly, the map sampling points are selected based on the random number generator according to the following process:

[0069] Step 1: Set the number of map sampling points j = 1;

[0070] Step 2: Set the joint angle i = 1;

[0071] Step 3: Use the random number generator rand() to randomly generate joint angles within the limits of the i-th joint angle of the robot. The expression is:

[0072] θ i =rand()%(θ i max -θ i min )+θ i min

[0073] Among them, θ i max ,θ i min They represent the maximum and minimum limits of the i-th joint angle of the robot arm, respectively, and % represents the remainder operation;

[0074] Step 4: Initialize a k-dimensional zero vector q j , represents the configuration of the k-degree-of-freedom manipulator, and θ i Assign to q j The expression for the elements at the corresponding positions is:

[0075] q j [i] = θ i

[0076] Among them, q j [i] represents vector q j The element at position i;

[0077] Step 5, let i=i+1, if i≤k, return to step 3;

[0078] Step 6: Get a set of robot arm configurations q j =(θ1,θ2,…,θ k ), the robot arm configuration q j As a sampling point of the landmark map, add it to the map sampling point set P;

[0079] Step 7, let j = j + 1;

[0080] Step 8: Set the maximum number of sampling points of the map to M. If j≤M, return to step 2, otherwise skip this step;

[0081] Step 9: Get the map sampling point set P:

[0082] P = {q1,q2,...,q M}

[0083] Among them, the selection of M is set according to the task requirements.

[0084] Step 102, designing a method for representing obstacles in a roadmap based on mapping of joint space coordinates and Cartesian space robot end coordinates;

[0085] Specifically, take out all the sampling points q=(θ1,θ2,...,θ k ), perform collision detection on all sampling points in turn, filter out all sampling points that collide with the robot itself or obstacles, use set form to represent the sampling points that collide with obstacles, and obtain the set of sampling points that collide with obstacles in the working environment:

[0086] S={q1,q2,...,q N}

[0087] Where S represents the set of sampling points where the robot arm collides with obstacles, q i (i=1, 2, ..., N) represents the sampling points in the set that collide with obstacles in the working environment. The sampling point set S is used to represent the obstacles in the working scene in the roadmap.

[0088] Step 103, based on the known state change information of obstacles in the scene, a multi-type sampling point connection strategy is designed to construct a new landmark map, and a landmark map containing all the state change information of the obstacles is obtained;

[0089] Specifically, firstly, the obstacles in the workspace are divided into fixed obstacles and variable obstacles. For variable obstacles, all variable states of the obstacles are represented in the workspace;

[0090] Secondly, the existing configuration set S of the robot arm colliding with obstacles is divided into sets S1 and S2 according to the collision with fixed obstacles and the collision with variable obstacles. S2 represents the configuration set of the robot arm colliding with various possible states of variable obstacles:

[0091]

[0092] in It represents the configuration in which the robot arm collides with variable obstacles A, B, etc. in states a, b, etc.;

[0093] Finally, various types of sampling points in the roadmap are connected according to the following process:

[0094] Step 1: The obtained map sampling point set P is divided into a collision point set S that collides with obstacles and a safe sampling point set P0 that does not collide with obstacles. The collision point set S is divided into a dangerous sampling point set S1 that collides with fixed obstacles and a threat sampling point set S2 that collides with variable obstacles.

[0095] Step 2: Delete the dangerous sampling point set S1 from the roadmap, and do not participate in the connection of sampling points in the roadmap. Let the remaining sampling point set in the roadmap be R;

[0096] Step 3: Let the number of sampling points in the road sign map i = 1, and the total number of sampling points in the sampling point set R be r;

[0097] Step 4: Calculate the sampling point q i The distance to all sampling points in the sampling point set R is expressed as:

[0098]

[0099] Among them, k is the degree of freedom of the robot arm, q j is the sampling point in set R, d j is the sampling point q i With sampling point q j The distance between the sampling points q i The distances from all sampling points in the sampling point set R are calculated using the set D i To characterize;

[0100] Step 5, in set D i Select the distance sampling point q i The nearest h sampling points are connected, and the weight of the connected edge is set to the distance between the sampling points. If the sampling point to be connected belongs to the safe sampling point set P0, the corresponding connected edge is defined as a reliable edge. If the sampling point to be connected belongs to the threat sampling point set S2, the corresponding connected edge is defined as an unreliable edge.

[0101] Step 6, let i = i + 1, if i ≤ r, return to step 4, otherwise skip this step;

[0102] Step 7: Get the landmark map G.

[0103] Step 104, based on the obtained landmark map, a landmark map search strategy for variable scenarios is designed based on the A* algorithm to complete the robot arm path search process.

[0104] Specifically, first, the current status of the variable obstacles is obtained from the working environment;

[0105] Secondly, the sampling point set S2' representing all states of the variable obstacle in the threat sampling point set S2 is taken out and matched with the current state of the variable obstacle. The sampling point q matching the current state of the variable obstacle is found in the sampling point set S2' representing all states of the variable obstacle. danger ;

[0106] Then, the sampling point q danger Converted into dangerous sampling points, all other sampling points in the threat sampling point set S2 are converted into safe sampling points, so that the threat sampling point set S2 in the road sign map G is converted into dangerous sampling points and safe sampling point sets;

[0107] Then, the unreliable edges connected to the dangerous sampling points are converted into dangerous edges, and the weights are set to +∞. The unreliable edges not connected to the dangerous sampling points are converted into reliable edges, and the weights are not changed.

[0108] Finally, the A* algorithm is used to search for the path with the smallest sum of edge weights from the starting point to the end point in the new landmark map to complete the robot's rapid obstacle avoidance path planning.

[0109] According to the above method provided by the embodiment of the present invention, a simulation experiment is performed on a four-degree-of-freedom robotic arm. Figure 2 , which is a four-degree-of-freedom robotic arm model. The DH parameters of the robotic arm are shown in Table 1.

[0110] Table 1 DH parameters of the robot arm

[0111]

[0112] The C++ language was used to build a simulation environment using the Qt development framework for experimental verification. The number of initial sampling points was selected as 1000, and the experimental operation environment was a personal computer with an Intel(R) Core(TM) i5-11400F@2.60GHz processor. This embodiment uses a two-dimensional plane to demonstrate the effectiveness of the improved PRM algorithm. Please refer to Figure 3 , Figure 4 . Figure 3 This is a schematic diagram of path planning after the threat obstacle is transformed into a fixed obstacle. It can be seen from the figure that the shortest path bypasses the threat obstacle; Figure 4 Because the threat obstacle has not become a fixed obstacle, it is not considered as an obstacle in the path planning process. From the planning results, it can be seen that the shortest path passes through the threat obstacle.

[0113] Select the following two configurations q start ,q end The starting and target configurations of this embodiment are configured, and obstacles that affect the operation of the robot arm are added in the robot arm motion space, including fixed obstacles and variable obstacles.

[0114] Starting configuration q start : [180°, -21°, 46°, 0°];

[0115] Target configuration q end : [120°, -21°, 46°, 0°];

[0116] After changing the state of the variable obstacle from A to B, the experimental results are shown in the following table.

[0117] Table 2 Obstacle avoidance path planning results

[0118]

[0119]

[0120] Due to the change of obstacle state, the result of obstacle avoidance path planning of the robot arm also changes. The planning result when the obstacle state is B is subjected to collision detection in the working environment where the obstacle state is A. When the robot arm configuration moves from [180°, -21°, 46°, 0°] to [168°, -21°, 48°, -176°], it collides with the obstacle in state A, while no collision occurs with the obstacle in state B during the whole process, which proves the effectiveness of the algorithm.

[0121] Using the RRT obstacle avoidance path planning method for q start ,q end Planning is carried out and the planning time is counted. The experimental results are shown in the following table.

[0122] Table 3 Comparison of planning time between RRT algorithm and improved PRM algorithm

[0123]

[0124] Under the experimental conditions, the improved planning algorithm reduced the planning time by 32.19% compared with the traditional RRT obstacle avoidance path planning method, proving the high efficiency of the algorithm.

[0125] The technical solution of the embodiment of the present invention has the following beneficial effects:

[0126] After representing variable obstacles in the landmark map, when the obstacle status in the robot arm's working scene changes, there is no need to spend a lot of time to reconstruct the landmark map. The shortest path can be quickly searched directly in the constructed landmark map, and the searched path meets the collision-free requirements.

[0127] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0128] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A method for fast obstacle avoidance path planning of a robotic arm for variable scenarios, characterized in that: The method comprises: Step S1 represents the configuration of the robot arm in the form of joint space coordinates, and uses a map sampling point selection strategy based on a random number generator to obtain a map sampling point set; Step S2 designs a method for representing obstacles in the roadmap based on the mapping of joint space coordinates and Cartesian space robot end coordinates; Step S3 designs a multi-type sampling point connection strategy to construct a new landmark map based on the known state change information of obstacles in the scene, and obtains a landmark map containing all the state change information of obstacles; In step S4, a landmark map search strategy for variable scenarios is designed based on the A* algorithm according to the obtained landmark map to complete the robot arm path search process.

2. The method according to claim 1, characterized in that The step S1 comprises: Step S1.1 represents a certain state of the k-degree-of-freedom robot in the workspace as a k-dimensional vector: q=(θ1,θ2,...,θ k ) Among them, θ i represents the joint angle of the i-th joint of the robot; Step S1.2 The map sampling point selection strategy based on the random number generator includes: Step S1.2.1: Set the number of map sampling points j = 1; Step S1.2.2: let joint angle i = 1; Step S1.2.3 uses the random number generator rand() to randomly generate joint angles within the limits of the i-th joint angle of the robot arm. The expression is: in, They represent the maximum and minimum limits of the i-th joint angle of the robot arm, respectively, and % represents the remainder operation; Step S1.2.4 Initialize a k-dimensional zero vector q j , represents the configuration of the k-degree-of-freedom manipulator, and θ i Assign to q j The expression for the elements at the corresponding positions is: q j [i]=θ i Among them, q j [i] represents vector q j The element at position i; Step S1.2.5: let i=i+1, if i≤k, return to step S1.2.3; Step S1.2.6 obtains a set of robot arm configurations q j =(θ1,θ2,...,θ k ), the robot arm configuration q j As a sampling point of the landmark map, add it to the map sampling point set P; Step S1.2.7: let j=j+1; Step S1.2.8: Let the maximum number of sampling points of the map be M. If j≤M, return to step S1.2.2, otherwise skip this step; Step S1.2.9 obtains the map sampling point set P: P={q1,q2,…,q M } Among them, the selection of M is set according to the task requirements.

3. The method according to claim 1, characterized in that The step S2 comprises: Take out all the sampling points q=(θ1,θ2,...,θ k ), perform collision detection on all sampling points in turn, filter out all sampling points that collide with the robot itself or obstacles, use set form to represent the sampling points that collide with obstacles, and obtain the set of sampling points that collide with obstacles in the working environment: S={q1,q2,...,q N } Where S represents the set of sampling points where the robot arm collides with obstacles, q i (i=1, 2, ..., N) represents the sampling points in the set that collide with obstacles in the working environment. The sampling point set S is used to represent the obstacles in the working scene in the roadmap.

4. The method according to claim 1, characterized in that: The step S3 comprises: Step S3.1: obstacles in the workspace are divided into fixed obstacles and variable obstacles. For variable obstacles, all variable states of the obstacles are represented in the workspace. Step S3.2 divides the existing configuration set S of the robot arm colliding with obstacles into sets S1 and S2 according to the collision with fixed obstacles and the collision with variable obstacles. S2 represents the configuration set of the robot arm colliding with various possible states of variable obstacles: in It represents the configuration in which the robot arm collides with variable obstacles A, B, etc. in states a, b, etc.; Step S3.3 The strategy for connecting multiple sampling points on the roadmap includes: Step S3.3.1: divide the obtained map sampling point set P into a collision point set S that collides with obstacles and a safe sampling point set P0 that does not collide with obstacles, and divide the collision point set S into a dangerous sampling point set S1 that collides with fixed obstacles and a threat sampling point set S2 that collides with variable obstacles; Step S3.3.2 deletes the dangerous sampling point set S1 from the road sign map, does not participate in the connection of sampling points in the road sign map, and sets the remaining sampling point set in the road sign map to R; Step S3.3.3: Let the number of sampling points in the road sign map i = 1, and the total number of sampling points in the sampling point set R be r; Step S3.3.4 Calculate the sampling point q i The distance to all sampling points in the sampling point set R is expressed as: Among them, k is the degree of freedom of the robot arm, q j is the sampling point in set R, d j is the sampling point q i With sampling point q j The distance between the sampling points q i The distances from all sampling points in the sampling point set R are calculated using the set D i To characterize; Step S3.3.5 In set D i Select the distance sampling point q i The nearest h sampling points are connected. The weight of the edge after connection is the distance between the sampling points. If the sampling point to be connected belongs to the safe sampling point set P0, the corresponding connected edge is defined as a reliable edge. If the sampling point to be connected belongs to the threat sampling point set S2, the corresponding connected edge is defined as an unreliable edge. Step S3.3.6: Let i = i + 1. If i ≤ r, return to step S3.3.4; otherwise, skip this step. Step S3.3.7 obtains the road landmark map G.

5. The method according to claim 1, characterized in that The step S4 comprises: Step S4.1 obtains the current state of the variable obstacle from the working environment; Step S4.2: Take out the sampling point set S2' representing all states of the variable obstacle from the threat sampling point set S2, match it with the current state of the variable obstacle, and find the sampling point q that matches the current state of the variable obstacle in the sampling point set S2' representing all states of the variable obstacle. danger ; Step S4.3: Set the sampling point q danger converted into dangerous sampling points, and all other sampling points in the threat sampling point set S2 are converted into safe sampling points; Step S4.4 converts the unreliable edges connected to the dangerous sampling points into dangerous edges, and sets the weights to +∞; converts the unreliable edges not connected to the dangerous sampling points into reliable edges, and does not change the weights; Step S4.5 uses the A* algorithm to search for the path with the smallest sum of edge weights from the starting point to the end point in the new landmark map, completing the robot arm's rapid obstacle avoidance path planning.

Images